SYNTHETIC HUMANITIES LAB / EXPERIMENT 03
Does the encounter have a history?
The same passage reaches a receiver after two different lead-ins.
Loading measured responses…
THE COMPARISON
Lead-in A contains six seconds of the synthetic recording. Lead-in B rearranges the same three two-second blocks. They have identical duration, sample distribution and total injected stimulation. Both are followed by exactly the same two-second passage, starting at exactly the same time.
For each lead-in we also run a quiet continuation, omitting the passage. This distinguishes an aftereffect from a change in how the passage is received. We repeat the four conditions with gaps of zero, half a second and two seconds.
| Earlier sound | What follows | Purpose |
|---|---|---|
| A | Shared passage | Response after A |
| B | Shared passage | Response after B |
| A | Quiet | Continuation of A without the passage |
| B | Quiet | Continuation of B without the passage |
104 simulations, eight new seeds (201–208), including eight independent duplicate runs. Seeds vary model background noise, not biological specimens. The original frozen connectome, neural dynamics and amplitude encoder are unchanged. These clock-aligned edits are diagnostic stimuli, not natural alternative performances.
RESPONSE / does history change reception?
| Gap before passage | Changed reception: first 0.5 s (primary) | Changed reception: full 2 s (secondary) | Lingering difference: first 0.5 s |
|---|
Primary population: direct JON postsynaptic partners, excluding injected cells. “Criterion met” means a repeatable temporal contrast by the rule fixed before running; it is not a significance test. “Not met” means unresolved at this sensitivity, not proven absence.
INSPECT A CONDITION
The changed-reception contrast subtracts the quiet continuation from each passage response, then compares those increments: ([B + passage] − [B + quiet]) − ([A + passage] − [A + quiet]). The labels here denote whole trial conditions; the subtractions compare spike counts.
| Contrast | Early mean (Hz/neuron) | Signs + / − / 0 | Early criterion | Full criterion |
|---|
Numerical criterion and time course
Using 100 ms bins, the RMS of the average contrast must exceed the RMS of its across-seed sample standard deviation, and the cosine similarity between the first four and last four repetitions must exceed 0.5. Both windows and all gaps were declared before running. The primary window is the first half-second.
| Seconds after passage onset | Changed-reception mean | Observed seed range |
|---|
Rates in Hz/neuron. Ranges describe the eight observed repetitions, not confidence intervals. No peak windows were selected after observing these data.
Listen to or download the exact diagnostic stimuli
PCM files are the exact samples encoded for these simulations. Cuts may interrupt spoken words; the experiment operates on timed audio blocks without language analysis. “Quiet” still includes the audible lead-in. These links open WAV files rather than start a coupled neural replay.
FUNCTIONAL IMPLICATION
READING / an affective interpretation
This rule-based interpretation receives only the measured contrast summary and gap lengths. It receives no poem text, waveform or performer identity. A history-dependent neural response would support a claim about how the modeled encounter unfolds, without identifying a subjective feeling.
Method and evidence
The quiet continuations estimate what each history would produce without the passage. The difference of probe increments tests non-additivity at the selected population readout. It does not locate a biological mechanism. Immediate pre-passage drive also differs between histories; any effect could come from ordinary membrane/spike dynamics or recurrence.
Within each gap the model starts identically and encounters the passage at the same absolute time. Across different gaps, absolute timing changes, so a delay trend would be descriptive rather than a fitted memory timescale. No learning or weight changes occur.
Full spikes are retained locally; public counts permit independent reconstruction. This recorded report does not start new online simulations.